Merge pull request #223 from PolicyEngine/calibration-fix-2

Calibration fix on SS target

Author: nikhilwoodruff

changelog_entry.yaml

@@ -0,0 +1,4 @@
+- bump: patch
+  changes:
+    added:
+    - SS HMRC calibration targets.

policyengine_uk_data/utils/loss.py

@@ -257,12 +257,11 @@ def pe_count(*variables):
 
     df["ons/uk_population"] = household_from_person(age >= 0)
 
-    # Filter to columns that exist in statistics (other targets added via target_names/target_values)
-    stats_for_period = statistics[
-        statistics.time_period == int(time_period)
-    ].set_index("name")
-    columns_in_stats = [c for c in df.columns if c in stats_for_period.index]
-    targets = stats_for_period.loc[columns_in_stats]
+    targets = (
+        statistics[statistics.time_period == int(time_period)]
+        .set_index("name")
+        .loc[df.columns]
+    )
 
     targets.value = np.select(
         [
@@ -328,6 +327,66 @@ def pe_count(*variables):
             target_values.append(row[variable + "_count"])
             target_names.append(name_count)
 
+    # HMRC Table 6.2 - Salary sacrifice income tax relief by tax rate
+    # This helps calibrate the distribution of SS users by income level
+    # 2023-24 values (£m): Basic £1,600, Higher £4,400, Additional £1,200
+    # Total IT relief from SS: £7,200m
+    # Use true counterfactual: IT relief = counterfactual IT - baseline IT
+    income_tax_baseline = sim.calculate("income_tax")
+    income_tax_cf = counterfactual_sim.calculate("income_tax", time_period)
+    it_relief = income_tax_cf - income_tax_baseline
+
+    # Get tax band from counterfactual adjusted net income (where SS is wages)
+    adjusted_net_income_cf = counterfactual_sim.calculate(
+        "adjusted_net_income", time_period
+    )
+    basic_rate_threshold = (
+        sim.tax_benefit_system.parameters.gov.hmrc.income_tax.rates.uk[
+            0
+        ].threshold(time_period)
+    )
+    higher_rate_threshold = (
+        sim.tax_benefit_system.parameters.gov.hmrc.income_tax.rates.uk[
+            1
+        ].threshold(time_period)
+    )
+    additional_rate_threshold = (
+        sim.tax_benefit_system.parameters.gov.hmrc.income_tax.rates.uk[
+            2
+        ].threshold(time_period)
+    )
+
+    # Determine tax band for each person based on counterfactual income
+    is_basic_rate = (adjusted_net_income_cf > basic_rate_threshold) & (
+        adjusted_net_income_cf <= higher_rate_threshold
+    )
+    is_higher_rate = (adjusted_net_income_cf > higher_rate_threshold) & (
+        adjusted_net_income_cf <= additional_rate_threshold
+    )
+    is_additional_rate = adjusted_net_income_cf > additional_rate_threshold
+
+    # Allocate the true IT relief to tax bands
+    ss_it_relief_basic = it_relief * is_basic_rate
+    ss_it_relief_higher = it_relief * is_higher_rate
+    ss_it_relief_additional = it_relief * is_additional_rate
+
+    df["hmrc/salary_sacrifice_it_relief_basic"] = household_from_person(
+        ss_it_relief_basic
+    )
+    df["hmrc/salary_sacrifice_it_relief_higher"] = household_from_person(
+        ss_it_relief_higher
+    )
+    df["hmrc/salary_sacrifice_it_relief_additional"] = household_from_person(
+        ss_it_relief_additional
+    )
+
+    # Total gross salary sacrifice contributions
+    # This is derived from the IT relief: £7.2bn IT relief at ~30% avg rate
+    # implies ~£24bn gross contributions (but we target the relief directly)
+    df["hmrc/salary_sacrifice_contributions"] = household_from_person(
+        ss_contributions
+    )
+
     # HMRC Table 6.2 - Salary sacrifice income tax relief by tax rate (2023-24)
     # https://assets.publishing.service.gov.uk/media/687a294e312ee8a5f0806b6d/Tables_6_1_and_6_2.csv
     # Values in £bn
@@ -352,6 +411,8 @@ def pe_count(*variables):
     target_names.append("hmrc/salary_sacrifice_contributions")
     target_values.append(SS_CONTRIBUTIONS_2024 * uprating_factor)
 
+    print(target_names[-4:], target_values[-4:])
+
     # Add two-child limit targets.
     child_is_affected = (
         sim.map_result(
@@ -476,6 +537,8 @@ def pe_count(*variables):
         ]
     )
 
+    combined_targets.to_csv("test.csv")
+
     return df, combined_targets.value